DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statement (IDS) was submitted on 1/09/2026. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Response to Amendment
This action is in response to the remarks filed on 1/09/2026. The amendments filed on 1/09/2026 are entered.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claims 1-3, 8-14, and 18-22 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by de Leeuw den Bouter, M.L., et al., (“Deep learning-based single image super-resolution for low-field MR brain images,” Scientific Reports. Vol 12(6362), 2022. P. 1-10) hereinafter de Leeuw den Bouter (see attached NPL reference of the office action of 7/09/2025 for citations).
Regarding claim 1, de Leeuw den Bouter teaches:
A method (abstract), comprising:
obtaining a first image of a brain with a low-field strength magnetic resonance imaging system, wherein the first image comprises a first resolution (page 8, Low-field MR image acquisition, brain image scans using low-field strength MR imaging);
obtaining a deep learning brain model based on high-field strength images (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model),
comprising high-field high-resolution images and high-field low-resolution images (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model, and also high-strength high resolution HR images are down sampled and converted with additions of noise to form high-field strength low-resolution images for training of the deep learning model. Therefore, the deep learning model is based upon both types of images. See particularly page 5, Dataset and training, paragraph 2)
wherein the deep learning brain model is configured to be applied by a neural network comprising a plurality of layers (pages 3-5, Convolutional neural network, neural network consists of a plurality of densely connected convolutional layers for deep-learning based methods; see also pages 5-8, Dataset and training and Results); and
applying the deep learning brain model to the first image to generate a second image of the brain, wherein the second image comprises a second resolution, and wherein the second resolution is greater than the first resolution (pages 8-9, Results and Discussion and conclusion, high resolution (HR) output images are obtained from the trained neural network which form second image of the brain at a higher second resolution based upon the input low resolution (LR) images to the trained neural network model; see also pages 3-8, Methods, Dataset and training).
Regarding claim 2, de Leeuw den Bouter teaches all of the limitations of claim 1. de Leeuw den Bouter further teaches:
wherein obtaining the deep learning brain model based on high-field strength images comprises obtaining a pre-trained model (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model, which forms a pre-trained model when utilized for application of an output high resolution image from a source input low resolution image).
Regarding claim 3, de Leeuw den Bouter teaches all of the limitations of claim 1. de Leeuw den Bouter further teaches:
wherein obtaining the deep learning brain model based on high-field strength images comprises:
accessing a high-field dataset comprising high-field strength high-resolution images and high-field strength low-resolution images (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model, and also high-strength high resolution HR images are down sampled and converted with additions of noise to form high-field strength low-resolution images for training of the deep learning model);
augmenting the high-field strength low-resolution images based on the high-field strength high-resolution images (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model, and also high-strength high resolution HR images are down sampled and converted with additions of noise to form high-field strength low-resolution images for training of the deep learning model. These low-resolution images are further added to the training process as augmented, downsampled, and gaussian noise augmented images); and
training the deep learning brain model based on the augmented high-field strength low-resolution images (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model, and also high-strength high resolution HR images are down sampled and converted with additions of noise to form high-field strength low-resolution images for training of the deep learning model.).
Regarding claim 8, de Leeuw den Bouter teaches all of the limitations of claim 3. de Leeuw den Bouter further teaches:
further comprising outputting the second image of the brain to a display (pages 8-9, Results and Discussion and conclusion, figure 5-6 show output images of an MRI scanner, which forms a high resolution deep learning model output image displayed).
Regarding claim 9, de Leeuw den Bouter teaches all of the limitations of claim 3. de Leeuw den Bouter further teaches:
further comprising displaying the second image of the brain (pages 8-9, Results and Discussion and conclusion, figure 5-6 show output images of an MRI scanner, which forms a high resolution deep learning model output image displayed).
Regarding claim 10, de Leeuw den Bouter teaches all of the limitations of claim 3. de Leeuw den Bouter further teaches:
wherein obtaining the first image of the brain with the low-field strength magnetic resonance imaging system comprises generating a low magnetic field strength of less than 100 mT (page 8, Low-field MR image acquisition, low-field MRI scanner is same as O’Reilly, and as further supported in reference [9] title, is a 50 mT scanner which is less than 100 mT), and
wherein the deep learning brain model is based on magnetic resonance imagining images obtained with a high magnetic field strength of more than 1 T (page 5, Dataset and training, high-field strength images are acquired using 1.5 T and 3 T MRI scanners which are more than 1 T).
Regarding claim 11, de Leeuw den Bouter teaches:
A system (abstract), comprising:
a processor; and a memory storing machine-readable instructions, wherein the processor is configured to execute the machine-readable instructions, and wherein the machine-readable instructions, when executed, implement a neural network (pages 3-8, Methods, Dataset and training, neural network processing includes processor and memory components for training and storing the neural network on a computer system as understood by one of ordinary skill in the art) configured to:
obtain a high-field strength magnetic resonance model comprising a plurality of layers (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model),
wherein the high-field strength magnetic resonance model is trained on high-field high-resolution images and high-field low-resolution images (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model, and also high-strength high resolution HR images are down sampled and converted with additions of noise to form high-field strength low-resolution images for training of the deep learning model. Therefore, the deep learning model is based upon both types of images. See particularly page 5, Dataset and training, paragraph 2);
receive data representative of a low-field image (page 8, Low-field image acquisition; see also pages 3-8, Methods, Dataset and training);
convert the low-field image to a higher resolution image based on the high-field strength magnetic resonance model (pages 8-9, Results and Discussion and conclusion, high resolution (HR) output images are obtained from the trained neural network which form second image of the brain at a higher second resolution based upon the input low resolution (LR) images to the trained neural network model; see also pages 3-8, Methods, Dataset and training); and
output the higher resolution image (pages 8-9, Results and Discussion and conclusion, high resolution (HR) output images are obtained from the trained neural network which form second image of the brain at a higher second resolution based upon the input low resolution (LR) images to the trained neural network model; see also pages 3-8, Methods, Dataset and training).
Regarding claim 12, de Leeuw den Bouter teaches all of the limitations of claim 11. de Leeuw den Bouter further teaches:
wherein the high-field strength magnetic resonance model comprises a pre-trained deep learning model (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model, which forms a pre-trained model when utilized for application of an output high resolution image from a source input low resolution image).
Regarding claim 13, de Leeuw den Bouter teaches all of the limitations of claim 12. de Leeuw den Bouter further teaches:
wherein the pre-trained deep learning model is trained with a high- field dataset comprising high-field strength high-resolution images and high-field strength low- resolution images (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model, and also high-strength high resolution HR images are down sampled and converted with additions of noise to form high-field strength low-resolution images for training of the deep learning model).
Regarding claim 14, de Leeuw den Bouter teaches all of the limitations of claim 11. de Leeuw den Bouter further teaches:
wherein the neural network is further configured to:
obtain a high-field dataset comprising high-field strength high-resolution images and high-field strength low-resolution images (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model, and also high-strength high resolution HR images are down sampled and converted with additions of noise to form high-field strength low-resolution images for training of the deep learning model);
augment the high-field strength low-resolution images based on the high-field strength high-resolution images (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model, and also high-strength high resolution HR images are down sampled and converted with additions of noise to form high-field strength low-resolution images for training of the deep learning model. These low-resolution images are further added to the training process as augmented, downsampled, and gaussian noise augmented images); and
train the high-field strength magnetic resonance model based on the augmented high- field strength low-resolution images (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model, and also high-strength high resolution HR images are down sampled and converted with additions of noise to form high-field strength low-resolution images for training of the deep learning model.).
Regarding claim 18, de Leeuw den Bouter teaches all of the limitations of claim 14. de Leeuw den Bouter further teaches:
wherein the high-field strength magnetic resonance model is trained with high-field images obtained at a high magnetic field strength exceeding 1 T (page 5, Dataset and training, high-field strength images are acquired using 1.5 T and 3 T MRI scanners which are more than 1 T).
Regarding claim 19, de Leeuw den Bouter teaches all of the limitations of claim 18. de Leeuw den Bouter further teaches:
wherein the low-field image is obtained at a low magnetic field strength of less than 0.3 T (page 8, Low-field MR image acquisition, low-field MRI scanner is same as O’Reilly, and as further supported in reference [9] title, is a 50 mT scanner which is less than 0.3 T).
Regarding claim 20, de Leeuw den Bouter teaches all of the limitations of claim 18. de Leeuw den Bouter further teaches:
wherein the low-field image is obtained at a low magnetic field strength of less than 100 mT (page 8, Low-field MR image acquisition, low-field MRI scanner is same as O’Reilly, and as further supported in reference [9] title, is a 50 mT scanner which is less than 100 mT).
Regarding claim 21, de Leeuw den Bouter teaches all of the limitations of claim 1. de Leeuw den Bouter further teaches:
further comprising augmenting the high-field low- resolution images with the high-field high-resolution images (pages 5-8, Dataset and training, high resolution high field strength images are utilized to generate the deep learning based neural network brain model, and also high-strength high resolution HR images are down sampled and converted with additions of noise to form high-field strength low-resolution images for training of the deep learning model. Therefore, the high-field low resolution images are augmented with the high-field high resolution images).
Regarding claim 22, de Leeuw den Bouter teaches all of the limitations of claim 1. de Leeuw den Bouter further teaches:
wherein the second resolution is greater than the first resolution by a factor of 2 (page 5, Dataset and training, paragraphs 1-2, both horizontal and vertical resolution provides a factor of two between the high resolution output of 128x128 pixels and the lower resolution of 64x64 pixels).
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 4-7 and 15-17 are rejected under 35 U.S.C. 103 as being unpatentable over de Leeuw den Bouter as applied to claims 3 or 14 above, and further in view of Schlemper et al. (U.S. Pub. No. 20200034998) hereinafter Schlemper.
Regarding claim 4, primary reference de Leeuw den Bouter teaches all of the limitations of claim 3. de Leeuw den Bouter further fails to teach:
further comprising performing transfer learning to fine-tune the deep learning brain model for the first image of the brain
However, the analogous art of Schlemper of deep learning techniques for magnetic resonance imaging reconstruction (abstract) teaches:
further comprising performing transfer learning to fine-tune the deep learning brain model for the first image of the brain ([0087]-[0088], little-low-field MRI data available for training and the “high-field” training dataset is larger than the low-field data, which is merely used for adapting the trained model using transfer learning; [0159]-[0161], high-field images are plentiful and form the basis of the dataset for training, with transfer learning using the smaller available dataset of low-field MRI data to fine tune the model; [0162]-[0163]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the low-field MR deep learning image resolution enhancement processing method of de Leeuw den Bouter to incorporate the transfer learning for fine-tuning the deep learning model generated from high-field data as taught by Schlemper because there is little low-field MRI data available for training and transfer learning enables high quality model generation using high-field dataset with fine tuning with the small availability of low-field MRI images (Schlemper, [0087]-[0088]; [0159]-[0161]). This leads to higher quality output images with the datasets currently available.
Regarding claim 5, primary reference de Leeuw den Bouter teaches all of the limitations of claim 3. de Leeuw den Bouter further fails to teach:
further comprising re-training at least one layer of the deep learning brain model with a low-field dataset
However, the analogous art of Schlemper of deep learning techniques for magnetic resonance imaging reconstruction (abstract) teaches:
further comprising re-training at least one layer of the deep learning brain model with a low-field dataset ([0087]-[0088], little-low-field MRI data available for training and the “high-field” training dataset is larger than the low-field data, which is merely used for adapting the trained model using transfer learning which forms a re-training of at least one layer of the model; [0159]-[0161], high-field images are plentiful and form the basis of the dataset for training, with transfer learning using the smaller available dataset of low-field MRI data to fine tune the model (re-training); [0162]-[0163]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the low-field MR deep learning image resolution enhancement processing method of de Leeuw den Bouter to incorporate the transfer learning with a low-field dataset for fine-tuning at least one layer of the deep learning model generated from high-field data as taught by Schlemper because there is little low-field MRI data available for training and transfer learning enables high quality model generation using high-field dataset with fine tuning with the small availability of low-field MRI images (Schlemper, [0087]-[0088]; [0159]-[0161]). This leads to higher quality output images with the datasets currently available.
Regarding claim 6, primary reference de Leeuw den Bouter teaches all of the limitations of claim 3. de Leeuw den Bouter further fails to teach:
further comprising re-training a subset of the layers of the deep learning brain model with a low-field dataset
However, the analogous art of Schlemper of deep learning techniques for magnetic resonance imaging reconstruction (abstract) teaches:
further comprising re-training a subset of the layers of the deep learning brain model with a low-field dataset ([0087]-[0088], little-low-field MRI data available for training and the “high-field” training dataset is larger than the low-field data, which is merely used for adapting the trained model using transfer learning which forms a re-training of at least one layer of the model; [0159]-[0161], high-field images are plentiful and form the basis of the dataset for training, with transfer learning using the smaller available dataset of low-field MRI data to fine tune the model (re-training); [0162]-[0163]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the low-field MR deep learning image resolution enhancement processing method of de Leeuw den Bouter to incorporate the transfer learning with a low-field dataset for fine-tuning at least one layer of the deep learning model generated from high-field data as taught by Schlemper because there is little low-field MRI data available for training and transfer learning enables high quality model generation using high-field dataset with fine tuning with the small availability of low-field MRI images (Schlemper, [0087]-[0088]; [0159]-[0161]). This leads to higher quality output images with the datasets currently available.
Regarding claim 7, primary reference de Leeuw den Bouter teaches all of the limitations of claim 3. de Leeuw den Bouter further fails to teach:
further comprising:
accessing a low-field dataset comprising low-field strength high-resolution images and low-field strength low-resolution images;
augmenting the low-field strength low-resolution images based on the low-field strength high-resolution images; and
re-training at least one layer of the deep learning brain model based on the augmented low-field strength low-resolution images
However, the analogous art of Schlemper of deep learning techniques for magnetic resonance imaging reconstruction (abstract) teaches:
further comprising:
accessing a low-field dataset comprising low-field strength high-resolution images and low-field strength low-resolution images ([0087]-[0088], mid-field strength MRI dataset forms a teaching to a “low-field” dataset as the field strength overlaps with the applicant’s disclosure of a low-field strength MRI system being permanent magnetic fields of less than 1.0 T (see paragraph [0055] of the applicant’s specification). Since these mid-field strength images are higher resolution than the low-field strength images of Schlemper, the Schlemper reference teaches to both high and low resolution low-field strength images in the combined “low-field” dataset; [0159]-[0161], mid-field and low-field images form the high and low resolution “low-field strength” image dataset; [0162]-[0163]);
augmenting the low-field strength low-resolution images based on the low-field strength high-resolution images ([0087]-[0088], mid-field strength MRI dataset forms a teaching to a “low-field” dataset as the field strength overlaps with the applicant’s disclosure of a low-field strength MRI system being permanent magnetic fields of less than 1.0 T (see paragraph [0055] of the applicant’s specification). Since these mid-field strength images are higher resolution than the low-field strength images of Schlemper, the Schlemper reference teaches to both high and low resolution low-field strength images in the combined “low-field” dataset. These mid-field strength images augment the low-field images in generating the transfer learning model with high-field strength image data; [0159]-[0161], mid-field and low-field images form the high and low resolution “low-field strength” image dataset, with the mid-field data augmenting in training a model using additionally the high-field data; [0162]-[0163]); and
re-training at least one layer of the deep learning brain model based on the augmented low-field strength low-resolution images ([0087]-[0088], little-low-field MRI data available for training and the “high-field” training dataset is larger than the low-field data, which is merely used for adapting the trained model using transfer learning which forms a re-training of at least one layer of the model; [0159]-[0161], high-field images are plentiful and form the basis of the dataset for training, with transfer learning using the smaller available dataset of low-field MRI data to fine tune the model (re-training); [0162]-[0163]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the low-field MR deep learning image resolution enhancement processing method of de Leeuw den Bouter to incorporate the transfer learning with a low-field dataset of higher and lower resolution data for fine-tuning the deep learning model generated from high-field data as taught by Schlemper because there is little low-field MRI data available for training and transfer learning enables high quality model generation using high-field dataset with fine tuning with the small availability of low-field MRI images (Schlemper, [0087]-[0088]; [0159]-[0161]). This leads to higher quality output images with the datasets currently available.
Regarding claim 15, primary reference de Leeuw den Bouter teaches all of the limitations of claim 14. de Leeuw den Bouter further fails to teach:
wherein the neural network is further configured to re-train at least one layer of the high-field strength magnetic resonance model with a low-field dataset
However, the analogous art of Schlemper of deep learning techniques for magnetic resonance imaging reconstruction (abstract) teaches:
wherein the neural network is further configured to re-train at least one layer of the high-field strength magnetic resonance model with a low-field dataset ([0087]-[0088], little-low-field MRI data available for training and the “high-field” training dataset is larger than the low-field data, which is merely used for adapting the trained model using transfer learning which forms a re-training of at least one layer of the model; [0159]-[0161], high-field images are plentiful and form the basis of the dataset for training, with transfer learning using the smaller available dataset of low-field MRI data to fine tune the model (re-training); [0162]-[0163]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the low-field MR deep learning image resolution enhancement processing system of de Leeuw den Bouter to incorporate the transfer learning with a low-field dataset for fine-tuning at least one layer of the deep learning model generated from high-field data as taught by Schlemper because there is little low-field MRI data available for training and transfer learning enables high quality model generation using high-field dataset with fine tuning with the small availability of low-field MRI images (Schlemper, [0087]-[0088]; [0159]-[0161]). This leads to higher quality output images with the datasets currently available.
Regarding claim 16, primary reference de Leeuw den Bouter teaches all of the limitations of claim 14. de Leeuw den Bouter further fails to teach:
wherein the neural network is further configured to:
obtain a low-field dataset comprising low-field strength high-resolution images and low- field strength low-resolution images;
augment the low-field strength low-resolution images based on the low-field strength high-resolution images; and
re-train at least one layer of the high-field strength magnetic resonance model based on the augmented low-field strength low-resolution images
However, the analogous art of Schlemper of deep learning techniques for magnetic resonance imaging reconstruction (abstract) teaches:
wherein the neural network is further configured to:
obtain a low-field dataset comprising low-field strength high-resolution images and low- field strength low-resolution images ([0087]-[0088], mid-field strength MRI dataset forms a teaching to a “low-field” dataset as the field strength overlaps with the applicant’s disclosure of a low-field strength MRI system being permanent magnetic fields of less than 1.0 T (see paragraph [0055] of the applicant’s specification). Since these mid-field strength images are higher resolution than the low-field strength images of Schlemper, the Schlemper reference teaches to both high and low resolution low-field strength images in the combined “low-field” dataset; [0159]-[0161], mid-field and low-field images form the high and low resolution “low-field strength” image dataset; [0162]-[0163]);
augment the low-field strength low-resolution images based on the low-field strength high-resolution images ([0087]-[0088], mid-field strength MRI dataset forms a teaching to a “low-field” dataset as the field strength overlaps with the applicant’s disclosure of a low-field strength MRI system being permanent magnetic fields of less than 1.0 T (see paragraph [0055] of the applicant’s specification). Since these mid-field strength images are higher resolution than the low-field strength images of Schlemper, the Schlemper reference teaches to both high and low resolution low-field strength images in the combined “low-field” dataset. These mid-field strength images augment the low-field images in generating the transfer learning model with high-field strength image data; [0159]-[0161], mid-field and low-field images form the high and low resolution “low-field strength” image dataset, with the mid-field data augmenting in training a model using additionally the high-field data; [0162]-[0163]); and
re-train at least one layer of the high-field strength magnetic resonance model based on the augmented low-field strength low-resolution images ([0087]-[0088], little-low-field MRI data available for training and the “high-field” training dataset is larger than the low-field data, which is merely used for adapting the trained model using transfer learning which forms a re-training of at least one layer of the model; [0159]-[0161], high-field images are plentiful and form the basis of the dataset for training, with transfer learning using the smaller available dataset of low-field MRI data to fine tune the model (re-training); [0162]-[0163]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the low-field MR deep learning image resolution enhancement processing system of de Leeuw den Bouter to incorporate the transfer learning with a low-field dataset of higher and lower resolution data for fine-tuning the deep learning model generated from high-field data as taught by Schlemper because there is little low-field MRI data available for training and transfer learning enables high quality model generation using high-field dataset with fine tuning with the small availability of low-field MRI images (Schlemper, [0087]-[0088]; [0159]-[0161]). This leads to higher quality output images with the datasets currently available.
Regarding claim 17, the combined references of de Leeuw den Bouter and Schlemper teach all of the limitations of claim 16. Primary reference de Leeuw den Bouter further fails to teach:
wherein the high-field dataset is larger than the low-field dataset
However, the analogous art of Schlemper of deep learning techniques for magnetic resonance imaging reconstruction (abstract) teaches:
wherein the high-field dataset is larger than the low-field dataset ([0087]-[0088], little-low-field MRI data available for training and the “high-field” training dataset is larger than the low-field data, which is merely used for adapting the trained model using transfer learning; [0159]-[0161], high-field images are plentiful and form the basis of the dataset for training, with transfer learning using the smaller available dataset of low-field MRI data to fine tune the model).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the low-field MR deep learning image resolution enhancement processing system of de Leeuw den Bouter and Schlemper to incorporate the larger high-field dataset as taught by Schlemper because there is little low-field MRI data available for training and transfer learning enables high quality model generation using high-field dataset with fine tuning with the small availability of low-field MRI images (Schlemper, [0087]-[0088]; [0159]-[0161]). This leads to higher quality output images with the datasets currently available.
Claim 23 is rejected under 35 U.S.C. 103 as being unpatentable over de Leeuw den Bouter as applied to claim 1 above, and further in view of Wong et al. (U.S. Pub. No. 20190192285) hereinafter Wong.
Regarding claim 23, primary reference de Leeuw den Bouter teaches all of the limitations of claim 1. de Leeuw den Bouter further fails to teach:
further comprising pre-processing the high-field strength images based on distortion correction, spatial normalization, or head masking
However, the analogous art of Wong of a machine learning technique used to train a classifier model for medical image processing (abstract) teaches:
further comprising pre-processing the training data images based on distortion correction, spatial normalization, or head masking ([0026], spatial normalization processing of training data images).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the low-field MR deep learning image resolution enhancement processing system of de Leeuw den Bouter to incorporate the spatial normalization processing as taught by Wong because it aligns the dataset across all images and eliminates potential inconsistences in coordinate systems and geometries (see also, Wong, [0026]). This improves the quality of the training dataset, leading to more accurate classification outputs with the final trained deep learning model.
Response to Arguments
Applicant's arguments filed 1/09/2026 have been fully considered but they are not persuasive. Responses to each of the applicant’s arguments are detailed below.
Regarding the applicant’s arguments on pages 7-8 of the remarks, the applicant argues that primary reference Bouter fails to teach to the present claim amendment in which the deep learning brain model is based upon high-field strength high-resolution images and high-field strength low-resolution images. The applicant argues that the reference fails to disclose high-field low resolution images. In the current rejections above, the reference is relied upon to teach to the model being “based on” a training dataset that includes both high-field high resolution images and high field down sampled low-resolution images (Bouter, page 5, Dataset and training, paragraph 2). These down sampled high field images teach to the broadest reasonable interpretation of the high field low resolution images as claimed. For these reasons, the applicant’s arguments directed to the independent claims are not persuasive.
Regarding the applicant’s arguments on page 8 of the remarks and directed to independent claims 1 and 11 and the Schlemper reference, the reference is not relied upon to teach to the associated limitations in the current rejections and therefore the arguments are not persuasive.
Regarding the applicant’s arguments on page 9 of the remarks and directed to dependent claim 7, the applicant argues that the Schlemper reference fails to teach to the associated resolutions as the mid-field strength images are only a higher resolution than the low-field images and not differentiated high and low resolutions as currently claimed. The applicant argues that the high and resolutions are particular resolution values, and Schlemper does not define the bounds of the captured resolutions. The particular ranges for high and resolutions as argued by the applicant correspond to paragraph [0064] of the applicant’s submitted specification which merely states exemplary resolutions (in the format of “can sometimes refer to”) and therefore does not refer to special definitions of the claim limitations in the specification. Any particular limited ranges of resolutions must be incorporated into the claim to be read into the broadest reasonable interpretation of the claim. Therefore, the current interpretation of the mid-field strength images being higher resolution than the low-field dataset is sufficient to teach to the high- and low-resolution images in the current rejections of claim 7 above.
For these reasons, the applicant’s arguments have been considered but are not persuasive.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Douglas et al. (U.S. Pub. No. 20200124691) teaches to imaging protocols that can be modified in real time using artificial intelligence processes. The method teaches to magnetic resonance imaging techniques including training data with high resolution imaging sequences.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SEAN A FRITH whose telephone number is (571)272-1292. The examiner can normally be reached M-Th 8:00-5:30 Second Fri 8:00-4:30.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Keith Raymond can be reached at 571-270-1790. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/SEAN A FRITH/Primary Examiner, Art Unit 3798
Prosecution Insights